Abstract:The outstanding performance of deep learning in various computer vision tasks motivated its application for medical image analysis, in particular, retinal fundus image analysis. It has been applied to a variety of tasks including diagnosis, detection and segmentation of pathologies in retinal fundus images. Many deep learning based techniques have been proposed to analyze retinal fundus images for automatic detection and diagnosis of macular degeneration and diabetic retinopathy. The automatic detection of diabetic retinopathy has the potential to prevent cases of vision loss and blindness by boosting the examination of diabetic patients. We carried out a comprehensive study of the latest deep learning techniques and their use in fundus image analysis. This paper presents the key concepts of deep learning relevant to diabetic retinopathy images analysis and reviews the latest deep learning based contributions in this area. We conclude the paper with a summary of the state-of-the-art, a critical discussion of open challenges and directions for future research.